Stroke is considered the second most common cause of dementia, but the magnitude of the risk posed by stroke has not been fully clarified. The aim of this study was to determine the long-term risk of developing dementia after stroke onset in a hospitalized cohort. We prospectively examined 185 nondemented patients aged > or = 60 years hospitalized with ischemic stroke and 241 age-matched nondemented controls without stroke from the same community using neurologic, neuropsychological, and functional assessments given annually. Using criteria modified from the DSM-III-R, we diagnosed incident dementia based on the annual examination findings. We used life-table methods to estimate incidence in the two groups, Kaplan-Meier analysis to determine the proportion surviving without dementia, and Cox proportional-hazards analysis to compute the relative risk (RR) of dementia after 1 to 4 years of follow-up. The incidence of dementia was 8.4 per 100 person-years in the stroke group and 1.3 per 100 person-years in the control group. After 52 months of follow-up, the cumulative proportion (+/- SE) surviving without dementia was 66.3 +/- 5.5% for stroke and 90.3 +/- 4.3% for control subjects. The RR of dementia associated with stroke compared with controls was 5.5 (95% CI, 2.5 to 11.1) after adjusting for demographic factors. Older age at stroke onset and fewer years of education were significant covariates, but sex and race were not. A low score on the Mini-Mental State Examination at baseline was a significant predictor when added to this model.(ABSTRACT TRUNCATED AT 250 WORDS)